Analytic Solutions for Real-Time Biosurveillance

Transcription

1 Executive Summary of the ISDS DTRA Consultancy Project The project was supported by contract #HDTRA1-15-C-0004 to ISDS from the Defense Threat Reduction Agency (DTRA) from November 2014 to March The (ISDS) fills the need for a practical forum and coordinating mechanism for collaboration among subject matter experts (SMEs) from stakeholder groups that may normally not interact but who, when brought together, enable innovative approaches to problems and solutions that are not possible by any one group alone. The objective of the Analytic Solutions for Real-Time Biosurveillance project was to advance analytic capabilities in real-time biosurveillance (BSV) by expediting next-generation solutions to currently intractable problems through focused consultancies that join end-user problem owners from civilian and military public health agencies with solution developers in academia, industry, and government. This approach helps to clarify key gaps in surveillance capabilities and to develop the requirements for knowledge management, algorithms, models, visualizations, and other solutions to address these gaps. This report summarizes the consultancies goals and outcomes and offers ideas for possible future consultancies should additional funding become available. Resources generated from these consultancies are available on the ISDS website at These consultancies generated two public health datasets that can be valuable to researchers working in public health. o A dataset containing variables from over a million records created specifically for public health practitioners examining risks for asthma exacerbations can be shared with solution developers through a Business Use Agreement (BUA) with Boston Public Health Commission. Information regarding the data set and BUA is available by contacting research@bphc.org. o The North Carolina NC Detect system released a dataset built from approximately 20,000 hospital visits captured via syndromic surveillance. Researchers can complete the data request form available at The form appears after the user goes through a required presentation. Call to Action: ISDS shall continue to advocate for opportunities to connect researchers and public health practitioners to both identify and implement sustainable improvements and processes in data health analytics. Connecting public health staff with additional training in data analytics will strengthen public health surveillance and lead to greater capabilities within state and local health departments. 1

2 Summary of consultancies #1-6. Consultancy #1: Asyndromic Cluster Detection, June 9-10, 2015 Problem Description: A syndrome cannot be created to identify every possible cluster of potential public health significance. A method is needed to identify clusters without pre-classification into syndromes. This could include clusters of signs or symptoms, clusters of place names (e.g. mentioning a specific restaurant), clusters of events (e.g. mentioning a specific fair, concert, etc.). Outcome: A paper was published in the Online journal of Public Health Informatics. The report from the meeting is available at: 0.pdf. Consultancy #2: Advancing the Utility of Infectious Disease Modeling for Public Health Practice, October 29-30, 2015 Problem Description: A syndrome cannot be created to identify every possible cluster of potential public health significance. A method is needed to identify clusters without pre-classification into syndromes. This could include clusters of signs or symptoms, clusters of place names (e.g. mentioning a specific restaurant), clusters of events (e.g. mentioning a specific fair, concert, etc.). Outcome: For access to the dataset from the North Carolina NC Detect system, requestors can complete the data request form available at The form appears after the user goes through a required presentation. The data elements in the dataset are listed below. Arrival date and time Masked facility ID Age group Chief complaint ~ 30 Injected Clusters The report from the meeting is available at: 2

3 Consultancy #3: Models for Forecasting Asthma Exacerbations in Urban Environments, March 30-31, 2016 Goals: The Boston Public Health Commission (BPHC) wanted to Identify a practical model to forecast environmental conditions that are likely to result in asthma exacerbations; Define systems requirements to automate the processing notifications of issues (environmental and technical). Communicate environmental conditions and actions to a variety of stakeholders at different levels of the socio-ecological model. Outcome: A paper titled Cross-Disciplinary Consultancy to Enhance Predictions of Asthma Exacerbation Risk in Boston was published in the Online Journal of Public Health Informatics in December The manuscript can be accessed at: In addition, a final report from the meeting was written and is available at: A dataset was created specifically for this use case to be shared with solution developers through a Business Use Agreement (BUA) with Boston Public Health Commission. Information regarding the data set and BUA is available by contacting research@bphc.org. ISDS contacted the Boston Public Health Commission for an update and were told that the health department staff continue to explore funding with partners. A common cold syndrome from this work was developed and they monitor activity and the association with asthma. Information is shared with the BPHC asthma program. Consultancy #4: Reportable Disease Cluster Detection in the Context of Sporadic Adoption of PCR-based Diagnostic Tests, call on May 3, 2016 Goals: Dr. Greene (NYC-DOHMH) and colleagues proposed a strategy to address their problem of excessive signaling by making recent and historical data more comparable. Requested input on the proposal, particularly from statisticians, or alternative suggestions to ensure the validity of the solution before proceeding with implementation. 3

4 Outcome: Feedback from NYC - DOHMH: I think we did get what we needed from the 5/3 call I was mainly looking for experts to consider our proposal and identify limitations we had missed or suggest a different direction. Overall, I m reassured that we have a reasonable approach that has been vetted to some degree by biosurveillance experts. We also received some good ideas on the call (e.g., a role for change-point analysis). After the final consultancy meeting in March of 2017, ISDS contacted NYC - DOHMH again to inquire about any additional progress related to the consultancy. NYC - DOHMH replied with an stating, We recently finished modifying the SAS code to implement the changes described in the proposal. We are running the modified code in parallel with the original code and comparing the results. The main limitation that no one seems to have anticipated (or at least expressed on the call according to my notes) is that the described adjustments do not work well for sparse data (e.g., for a rare disease citywide, or for a common disease applied to a small geographic resolution) in this situation, the adjustments make the counts in the baseline period overly spiky, so to speak, and unreliable. I think with the sporadic adoption of PCR-based diagnostic tests, at least for the time being, we have unfortunately lost the ability to usefully apply the refined historical limits method to sparse reportable disease data. I believe this technical problem might still warrant further thinking by ISDS biostatisticians experts, as health departments in jurisdictions across the U.S. must be facing this issue. ISDS has contacted Dr. Howard Burkom at Johns Hopkins Applied Physics Laboratory for a recommendation of a biostatistics expert to assist NYC - DOHMH. ISDS has reached out to those contacts and is in the process of connecting those experts with NYC - DOHMH so they can discuss the issue of sparse data points. Consultancy #5: Assessing Risk for Emerging Arboviral Disease, June 14-15, 2016 Goal: To integrate laboratory, demographic, syndromic, vector, climate, and geographic data to better understand the areas at greatest risk for importation and local transmission of dengue, chikungunya or Zika viruses to determine when and where to implement public health response actions. 4

5 Outcome: A group named, Fight the Bite was formed and now has designated group space on the ISDS website so interested partners can share information to sustain and share the work started in the consultancy. The final report for the meeting can be obtained by contacting the Fight the Bite group on the ISDS website or ing syndromicsurveillance@azdhs.gov. Consultancy 6: Negation Processing in Free Text Emergency Department Data for Public Health Surveillance, January 19-20, 2017 Goal: To discuss existing approaches to and gaps in negation processing techniques and discuss requirements for solutions to address those gaps. Outcome: A workgroup was formed to submit a paper to the Online Journal of Public Health Informatics. Participants were surveyed and interested parties were connected with an organizer who volunteered to lead the effort. The work for this paper is in progress. 5

6 Final In-Person Meeting: Meta-Consultancy on Sustainable Health Practices, March 16, 2017 During the course of this project, a common theme discussed among participants is the issue of sustaining a novel surveillance practice or tool identified in grant funded research projects where a researcher works with a public health agency. Often, the funding only pertains to the time used for the research and release of the information, not for the time needed to implement the new work. Another issue that generated great discussion was the lack of opportunity to train existing health department staff in data analytics and the barriers faced when trying to recruit staff trained in informatics to public health positions which typically pay considerably less than the private sector. The two goals of the final meeting were to discuss how to build sustainable implementation of a newly identified public health practice/tool and to identify training opportunities in data analysis for public health staff. The next steps with this final meeting may involve a letter to the editor where we discuss the need for sustainable implementation practices, a webinar to share training data analytics training opportunities with public health partners, and organized discussions among funder (CDC, NLM, NIH) to stress the need to change the language of the research grants so resources may be allocated to the sustained implementation of a new process. Discussion Summary and Conclusions Technical Conventions Committee The ISDS group that initiated this DTRA project was the Technical Conventions Committee. There was discussion about continuing that group starting in April of 2017 with a possible rebranding and identification of new committee chairs to lead the group. The consultancy participants reviewed the previous mission, The purpose of the ISDS Technical Conventions Committee is to facilitate and expedite the development, evaluation, and implementation of technical methods for public health surveillance. Those interested in being a part of the committee will either vote to keep the previous mission or may decide to update the language. Training of public health staff With the ending of the SHINE fellowship from CDC, health departments are looking for other opportunities to train existing public health staff in data analytics and for new ways to engage informatics graduates in public health. The group discussed other possible fellowship opportunities: 6

7 The Science, Mathematics And Research for Transformation (SMART) Scholarship for Service Program established by the Department of Defense (DoD). Participants shared that this opportunity has ended and is no longer an option. Presidential Innovation ( Participants shared that this opportunity also ended, but there is no confirmation on the website. National Library of Medicine. This program is still funded and operating. ISDS will continue to work with partners to identify training opportunities and will post announcements on both the ISDS website and via social media. The group discussed possible outcomes from this final meeting and suggestions include a letter to editor titled, Enabling advances in public health surveillance, an ISDS webinar, and an organized discussion among possible funders such as CDC, National Library of Medicine, and the National Institutes of Health. A letter to the editor could include: o Summary of what was done o Worked, hasn t worked o Recommendations o Funding Strategies o Staffing Strategies o Development Strategies o Mike: Practical projects o Successes Requirement analysis Practical outcomes o Asyndromic datasets o Asthma datasets o Fight the Bite o UT datasets Toward the conclusion of the final consultancy, participants were asked to offer feedback regarding what benefit, if any, they received from the final meeting. The following are the list of responses with the speaker de-identified: o How to pursue funding o Crowdsourcing o Management ideas o Coordinating conferences o Public Health, research and industry Face to face interactions o Finding good facilitators-there is a need for this to make us more successful o Simple concept papers for dissemination 7

8 o Perceived legal barriers o Identify model language better data sharing o Impart message better to Health Departments o Better support for programmatic management o Have more Syndromic Surveillance in informatics track o Put faces to names o Better understanding successes o New insight in informatics in terms of direction and challenges o Ideas for partnerships and collaborations o Remind to think outside of the bureaucratic box o To hear other ways to do things o To consider just doing things now o Appreciate seeing everyone o See new opportunities and new energy o Can take advantage of this new energy o Can t lose sight of value of curating and messaging o Need to get new staff up to speed o Finally figuring out what happened to the different consultancies o Wonder if there s a low budget way to continue work o Want to make these more accessible o Had no idea of the complementary nature of PHII and NACCHO o See tangential relationships o Wake up call that even if you see new technology, the process of getting it implemented is onerous o See the different rates of operation amongst government, and Public Health and research o Saw the connection between the different consultancies o See opportunity for associations like us to fill gap between PH and research Implementing research that s proven Very little interest in piloting implementation Need motivation from the researchers to complete implementation Journals: need knowledge library for researchers to access One participant expressed frustration that many of the obstacles we discussed were not new and public health has been talking about them for years with little resolve. While some participants agreed that these are not new problems, the issues are complex and require actions on many partners to overcome them. The participants gave suggestions to frame this final report for DTRA as a call to action. The report could also be shared with other potential funders to in an effort to fund additional consultancies. Lastly, this final report can be a practical way to let people know the datasets created during the project exist. 8

9 Possible Future Consultancies If additional funding were to be available, ISDS and the group have identified a variety of other possible consultancies. 1. Opioid surveillance According to the Centers for Disease Control and Prevention (CDC), 91 people die every day in the U.S. due to an opioid overdose ( While many health departments have access to death certificates, there are numerous other valuable data sources that are critical in capturing the full spectrum of the opioid addiction epidemic. Some health departments have access to counts of administered naloxone doses, coroner data detailing the specifics drugs identified during autopsies, needle exchange programs, just to name just a few. This emerging problem has prompted public health practitioners to collaborate with new partners and analyze novel data sources. While many health departments are finding themselves in this new role, little documentation or guidance exists to guide public health in how to best access, analyze, report and share the data. Many researchers and public health practitioners have shared their surveillance work through published papers, poster and oral presentations at conferences, and through educational opportunities such as webinars and conferences calls. However, much of this work continues to be done in disjointed silos. Those who have excelled in conducting opioid addiction and overdose surveillance utilizing one data source are not always connected to others in their community with complementary sources, which when combined, could create a more robust, real-time common operating picture. 2. Antibiotic resistance An executive order from the White house in 2014 led the CDC to release a National Strategy for Combating Antibiotic Resistant Bacteria ( ) As more states include antibiotic resistance diagnoses in their reportable conditions list, there are more inquiries from state and local health department on how to best capture prevalence in their communities. Several novel ideas have been shared among individual meetings, but there is still an urgent need to document suggested surveillance strategies for state and local health officials. CDC is working to address the threat in four areas: Slow the development of resistant bacteria and prevent the spread of resistant infections. Strengthen national one-health surveillance efforts to combat resistance. 9

10 Advance development and use of rapid and innovative diagnostic tests for identification and characterization of resistant bacteria. Improve international collaboration and capacities for antibiotic resistance prevention, surveillance, control and antibiotic research and development. There is ample need for increased surveillance work in each of these four areas. 3. Chronic conditions While many of the large advances in real time surveillance have centered around response to infectious diseases such as a Ebola and Zika, chronic conditions still remain the largest cause of morbidity and mortality in the United States. There is a need for more sharing of evidence based practices in conducting accurate surveillance of the chronic issues that plague our communities. As with the consultancy we hosted on asthma exacerbations, there is an abundance of opportunities to have similar collaborations on other chronic disease issues. As more health care providers move to the use of electronic health records, public health has new methods to capture a more robust picture of health. 4. Sharing surveillance summaries with emergency response partners As surveillance systems have become more robust in the last 15 years and health departments have generated more reports with their data, there is still room for improvement in reaching the correct target audiences to share the information. More guidance and coaching are needed on how to reach out to non-traditional partners and make connections for that information is shared across all sectors. Possible future partners for additional consultancies To continue the work and collaborations started with these consultancies and with the hope of hosting additional consultancies, participants were asked to share ideas about possible funding partners. The group generated a diverse list of commercial and public health partners. ISDS will reach out to these groups over the coming weeks to inquire about possible opportunities. Pharmacies (CVS, Walgreens, Rite Aid) First Watch National Science Foundation Skoll Global Threats Fund, RWJ, Sloan Foundation Google, Facebook, Microsoft (Foundation), Amazon, IBM All Scripts, Next Gen, Optima Siemens, Kaiser, Tabloid R, SAS 10

11 MIRTH Rhapsody Orion o Companies that hire public health informaticians for consulting (Booz Allen) For more information about these consultancies or to learn how to connect with ISDS, please visit or the ISDS Executive Director, Shandy Dearth, at sdearth@syndromic.org. 11